Redmine migration: BioPerl / BioSQL - InterPro Optional IDs not parsed

[cjfields]

Added by johnraekwon empty about 9 years ago. Updated over 7 years ago.

This bug was first posted on the BioSQL-l, and it was requested that I post it here.

The problem is that the InterPro IDs (aka optional ID) are not entered into the biosql database when using the script load_seqdatabase.pl

This information should probably rightfully go into the database field: dbxref_qualifier_value. In fact, this table is not utilized at all.

The reply to my original post on BioSQL-l is shown here:

The problem here is that Bioperl-db (the persistence mapper between BioSQL and BioPerl) loses the optional_id property of Bio::Annotation::DBLink objects.

Moreover, the dbxref table in BioSQL doesn't actually provide for the opportunity to store two identifiers (or accessions) for one db_xref, so storing this bit of information is actually not as straightforward as one might wish b/c it would need to go into the dbxref_qualifier_value table, and I would not be surprised if the other Bio* projects with a mapping to BioSQL don't store or retrieve this either (though it'd be good to hear if anyone does).

Here are a couple of ideas for how this issue might be addressed.

• Write a Bio::Seq::BaseSeqProcessor-derived object that for every incoming sequence massages all Interpro links to either substitute the primary_id with the optional_id, or to add a second DBLink annotation with the optional_id of the original one as its primary_id. (pros: relatively easy, entirely under your control; cons: you either lose the primary_id now, or have two dbxref annotations for each of the original ones.)

• Add a column to the dbxref table, and code to Bioperl-db, that store, de/serialize the extra ID. (pros: not losing or duplicating any data; cons: change is significant in terms of schema stability, requires new release, depends on implementation in Bioperl-db, necessitates update of all other Bio* language bindings)

• De/serialize the optional_id as an entry in the dbxref_qualifier_value table. (pros: technically it's the Right Way as that's what the table was intended for; cons: implementing in Bioperl-db is more involved as we now need to transform an object property to a child object and back)

So I'd say this is a bug in Bioperl-db in that the dbxref_qualifier_value table isn't utilized here. Would you mind filing it? In the meantime, if you just need something that works, you could try the first of the above ideas.

-hilmar

On Jul 3, 2009, at 7:17 PM, John LaCava wrote:

Hi all,

Tried this on the BioPerl-l but seemed to make sense to try here as well.

I am trying to use the BioPerl-db script:

"load_seqdatabase.pl" to parse a SwissProt ".dat" file (Yeast.dat, this is the yeast proteome with annotations etc.).

The particular entry I am interested is the InterPro optional ID, which is the domain name.

I have put a short stub up which displays the 4 pieces of info I want to parse into my data base.
That can be found here:

http://github.com/johnraekwon/BioPerl---BioSQL---InterPro-Optional-IDs/tree/master

You can see that near the bottom, we get the optional ID:
$protein_ids->{interpro_domain} = $dblink->{optional_id};

I do not think the bioperl script load_seqdatabase.pl retrieves this information. At least, I cannot find it in the db built from parsing a test .dat file.
I would like some help figuring out:

1. WHY doesn't it retrieve this information, since it seems to be parsing "all" annotations...
2. HOW might I edit the script to include this particular annotation of interest in the info it passes to my db (biosql)

I am a bit out of my depth on this, and so, any help is appreciated.

Cheers,
John

[cjfields]

Updated by Jason Stajich over 7 years ago

Priority changed from Urgent to Normal